Detailed characterization of deep level defects in InGaN Schottky diodes by optical and thermal deep level spectroscopies Cite as: Appl. Phys. Lett. 99, 092109 (2011); https://doi.org/10.1063/1.3631678 Submitted: 21 June 2011 . Accepted: 05 August 2011 . Published Online: 01 September 2011 Emre Gür, Zeng Zhang, Sriram Krishnamoorthy, S. Rajan, and S. A. Ringel ARTICLES YOU MAY BE INTERESTED IN Polarization-engineered GaN/InGaN/GaN tunnel diodes Applied Physics Letters 97, 203502 (2010); https://doi.org/10.1063/1.3517481 Luminescence properties of defects in GaN Journal of Applied Physics 97, 061301 (2005); https://doi.org/10.1063/1.1868059 Spontaneous emission of localized excitons in InGaN single and multiquantum well structures Applied Physics Letters 69, 4188 (1996); https://doi.org/10.1063/1.116981 Appl. Phys. Lett. 99, 092109 (2011); https://doi.org/10.1063/1.3631678 99, 092109 © 2011 American Institute of Physics. APPLIED PHYSICS LETTERS 99, 092109 (2011) Detailed characterization of deep level defects in InGaN Schottky diodes by optical and thermal deep level spectroscopies Emre Gu¨r,1,2,a) Zeng Zhang,2 Sriram Krishnamoorthy,2 S. Rajan,2 and S. A. Ringel2 1Faculty of Science, Department of Physics, Atatu¨rk University, Erzurum 25240, Turkey 2Department of Electrical and Computer Engineering, The Ohio State University, Columbus, Ohio 43210, USA (Received 21 June 2011; accepted 5 August 2011; published online 1 September 2011; publisher error corrected 15 September 2011) Schottky diode properties of semitransparent Ag(4 nm)/Au(4 nm) metal stack on In0.2Ga0.8N were investigated and defect characterization was performed using capacitance deep level transient (DLTS) and optical spectroscopy (DLOS). DLTS measurements made on the In0.2Ga0.8N Schottky diodes, which displayed a barrier height of 0.66 eV, revealed the presence of two deep levels located at Ec-0.39 eV and Ec-0.89 eV with nearly identical concentrations of 1.2 Â 1015 cmÀ3. Three deeper defect levels were observed by DLOS at Ec-1.45 eV, Ec-1.76 eV, and Ec-2.50 eV with concentrations of 1.3 Â 1015cmÀ3, 3.2 Â 1015cmÀ3, and 6.1 Â 1016 cmÀ3, respectively. The latter, with its high trap concentration and energy position lying 0.4 eV above the valance band, suggests a possible role in compensation of carrier concentration, whereas the mid-gap positions of the other two levels imply that they will be important recombination-generation centers VC 2011 American Institute of Physics. [doi:10.1063/1.3631678] 7,8 There has been great interest in GaN and related alloys quality InxGa1ÀxN layers, larger amount of surface defects due to their superior properties such as high breakdown field, compared to GaN,9 and the presence of surface accumulation good thermal conductivity, and high electron saturation ve- layers.7 This study focuses on the development of high qual- 1 locity. While AlxGa1ÀxN alloys are important for power ity In0.2Ga0.8N SDs and subsequent electrical and trap char- devices and deep ultraviolet emitters, InxGa1ÀxN alloys with acterization using both DLTS and DLOS measurements. bandgaps ranging from 0.7 to 3.4 eV covering the entire visi- Plasma-assisted molecular beam epitaxy (PAMBE) was ble spectrum are of interest for many future optoelectronic used to grow 300 nm thick, unintentionally doped (uid) and energy applications. In addition, high In composition InxGa1ÀxN layers on an n-type GaN template (Kyma, carrier 18 À3 8 À2 InxGa1ÀxN alloys also possess significant potential for high density 1 Â 10 cm and dislocation density 10 cm ). speed device applications due to their small effective mass PAMBE growth optimization was guided using a growth and high mobility.2 For all such device applications, device model developed for Ga-polar InGaN reported elsewhere.10 characteristics and performance are greatly impacted, and The In0.2Ga0.8N composition was confirmed by high resolu- degraded, by the presence of variety of growth related tion triple-axis x-ray diffraction measurements. After growth, defects. Recently, it has been shown that the defect states the sample was cleaned by dipping into the HCl in order to formed either during the growth process or as a result of de- get rid of the In droplets and then standard organic cleaning vice operation tend to act as non-radiative recombination process followed by acetone/methanol/isopropanol. Optical centers and/or create leakage pathways that degrade vertical (contact) lithography was used to define patterns after which conduction in InxGa1ÀxN-based quantum well light emitting O2 ashing and HCl:DI (1:10) were used to remove photoresist diodes (LEDs) and blu-ray laser diodes.3,4 However, the cur- residual and native oxides, respectively. An electron beam rent state of knowledge about the electronic properties of evaporated Ag(4 nm)/Au(4 nm) metal stack was used as a defects in InxGa1ÀxN is very limited and has not been well semitransparent Schottky contact to allow penetration of explored. A contributing reason is the difficulty in obtaining monochromatic light used in DLOS measurements. Devices high-quality InGaN Schottky diodes (SDs) that would enable were isolated using Cl2–based reactive ion etching. Finally, a quantitative investigation of electronic defect properties by Ti:Al:Ni:Au metal stack was evaporated for ohmic contacts capacitance-based transient techniques, in particular deep on the nþ-GaN template. level transient (DLTS) and optical spectroscopy (DLOS). To Fig. 1 shows the electrical properties of the Au/Ag/ date, there have been sparse reports related to the SD proper- In0.2Ga0.8N SD by current voltage (I-V) and capacitance- ties of InxGa1ÀxN. These studies have generally revealed voltage (C-V) measurements measured at 300 K. Almost poor quality diodes, even for low In composition alloys com- three orders of magnitude rectification factor is evident from pared to SDs those on GaN,5,6 which have been attributed to Fig. 1(a). Also, the reverse current density of the Au/Ag/ high background doping and difficulties in obtaining high In0.2Ga0.8N SC shows field dependence, i.e., no saturation, which might be either due to the voltage dependency of the barrier height (perhaps due to image force lowering), and/or a)Author to whom correspondence should be addressed. Electronic tunneling, or is due to generation of current via bandgap addresses: [email protected] and [email protected]. Present states in the depletion region.11 From the I-V analysis, ideal- address: 205 Dreese Laboratory, Department of Electrical and Computer Engineering, Ohio State University, 2015 Neil Avenue, Columbus, Ohio ity factor, barrier height, and saturation current density 43210-1272, USA. Tel.: þ1 614 247 7111. FAX: þ1 614 292 9562. parameters were determined to be n ¼ 1.34, Ub ¼ 0.66 eV, 0003-6951/2011/99(9)/092109/3/$30.00 99, 092109-1 VC 2011 American Institute of Physics 092109-2 Gu¨r et al. Appl. Phys. Lett. 99, 092109 (2011) shown in Fig. 2. The energy levels and capture cross section values of these traps were determined by Arrhenius analysis to be Ec-0.39 eV, 1.24 Â 10À16 cm2 and Ec-0.89 eV, 1.1 Â 10À13 cm2, respectively, with concentrations of 1.9 Â 1015 cmÀ3 and 2.8 Â 1015 cmÀ3. Although in the fig- ure the peak heights appear quite similar, the difference in actual concentrations comes from the lambda effect correc- tion to the concentration values. Since depletion depth at V ¼ 0V,40 nm as shown in Fig. 1(b), is large in compari- son to the distance where the traps are never filled (for the deeper level, the value is only 2 nm), this distance is neglected. The details of the correction method can be found in Ref. 13. While reports of DLTS-determined traps in InGaN are sparse, Soh et al. reported a deep level that appears to be close to the Ec – 0.39 eV trap observed here with similar cap- ture cross section and energy values (1.2 Â 10À16 cm2, 0.40 eV, respectively);14 however this trap, which in that work was ascribed to In segregation, was observed for 14% InGaN material, and thus its correlation with the state observed here is only speculative. In order to probe the defect states located in the remain- der of the 2.9 eV bandgap, DLOS measurements were per- formed using a monochromatized quartz-tungsten halogen (QTH) and Xe lamp with photon energies scanned between 0.5-2.0 eV and 1.2-3.6 eV, respectively. Complete DLOS details have been provided elsewhere.15–18 Both QTH and Xe FIG. 1. (Color online) (a) I-V and (b) C-V characteristics of the Au/Ag/ lamp DLOS measurements were performed at reverse bias of In0.2Ga0.8N Schottky diode. –0.5 V and filling bias of 0 V with filling pulse time of 10 s. The measurement time used was 200 s, which commenced af- and 2 Â 10À5 A/cm2, respectively, by modeling the meas- ter an 80 s thermal settling time to avoid signals from thermal ured data as a standard Schottky barrier following simple emission from deep levels. The steady state photo-capacitance thermionic emission.11 The relatively good ideality factor (SSPC) for each detected state provides individual trap con- (i.e., close to 1) compared with prior reports of InGaN SDs centrations and a full photo-capacitance transient analysis supports that thermionic emission is the dominant current enables extraction of the optical cross section from each transport mechanism.5,6 The characteristic energy of the detected level, from which precise energy level values are 15–18 Schottky barrier, qE00, was calculated to be 10.2 meV, which obtained.